Angiostatin Overexpression in Morris Hepatoma Results in Decreased Tumor Growth but Increased Perfusion and Vascularization

Kerstin Schmidt, MS1–3; Johannes Hoffend MD3; Annette Altmann, PhD2,3; Ludwig G. Strauss, MD, PhD2; Antonia Dimitrakopoulou-Strauss, MD2; Britta Engelhardt, MD1; Dirk Koczan, MD4;Jo¨rg Peter, PhD5; Thomas J. Dengler, MD6; Walter Mier, PhD2; Michael Eisenhut, PhD7; Uwe Haberkorn, MD, PhD2,3; and Ralf Kinscherf, PhD1

1Department of Anatomy and Cell Biology III, University of Heidelberg, INF 307, Heidelberg, Germany; 2Clinical Cooperation Unit Nuclear Medicine, DKFZ and University of Heidelberg, INF 280, Heidelberg, Germany; 3Department of Nuclear Medicine, University of Heidelberg, INF 400, Heidelberg, Germany; 4Department of Immunology, University of Rostock, Rostock, Germany; 5Department of Biophysics and Medical Radiation Physics, DKFZ, INF 280, Heidelberg, Germany; 6Department of Internal Medicine III, INF 410, University of Heidelberg, Germany; and 7Department of Radiopharmaceutical Chemistry, German Cancer Research Centre, Heidelberg, Germany

Key Words: ; gene therapy; PET; ; Growth of malignant tumors is dependent on sufficient perfusion supply. Thus, inhibition of tumor angiogenesis is emerging as a J Nucl Med 2006; 47:543–551 promising target in the treatment of malignancies. Human angiostatin (hANG) is one of the most potent inhibitors of endo- thelial cell proliferation, angiogenesis, and tumor growth in vivo. However, its mechanisms operating in vivo are not well un- derstood. Methods: To obtain more information about functional Tumor growth is angiogenesis dependent (1). Thus, changes in the angiogenic process, we established Morris hepa- inhibition of tumor angiogenesis is emerging as a promis- toma (MH3924A) cell lines expressing hANG (hANG-MH3924A). ing target in the treatment of malignancies as well as for the The effects of hANG expression on proliferation and of human umbilical vein endothelial cells (HUVECs) were mea- prevention of cancer recurrence or metastasis (2). Novel sured in coculture experiments in vitro. To evaluate changes in strategies, including endogenous antiangiogenic agents, 15 68 tumor perfusion and blood volume, H2 O and Ga-DOTA- which starve tumors by disturbing blood supply, have been (DOTA is 1,4,7,10-tetraazacyclododecane-N,N,N9 $,N%- shown to be effective in preclinical models. The advantages tetraacetic acid) were used for PET studies in vivo. Additionally, of focusing on blood supply are prevention of drug resis- immunohistologic quantification of vascularization, apoptosis, tance and a highly targeted therapy of a variety of tumor and proliferation as well as gene array analyses were performed. entities (3). Human angiostatin (hANG), a cleavage product Results: Our in vitro experiments demonstrate reduced prolif- eration and increased apoptosis in HUVECs when being co- of plasminogen, is one of the most potent antiangiogenic cultured with hANG-MH3924A. In support, tumor growth of agents currently known (4). ANG exhibits profound inhi- hANG-MH3924A is diminished by 95% in vivo. However, tumor bition of endothelial cell proliferation and migration in perfusion and blood volume are increased in hANG-MH3924A vitro and angiogenesis in vivo (5). Systemic administration corresponding to an increased microvessel density. Further- of recombinant ANG inhibited angiogenesis and increased more, hANG-transfected tumors show changes in expression the apoptotic rate in several tumor models followed by of genes related to apoptosis, stress, signal transduction, and tumor regression (6). The formation of ANG implicates . Conclusion: hANG expression leads to inhibition of tumor growth, increased apoptosis, and changes in the ex- or serine proteinases produced by tumor pression of multiple genes involved in stress reactions, signal cells (7) or by tumor-infiltrating macrophages (8). However, transduction, and apoptosis, which indicates a multifactorial re- the molecular mechanisms of ANG treatment in vivo remain action of tumors. An enhanced microvessel density is seen as ill defined. Therefore, monitoring of antiangiogenic ap- part of these reactions and is associated with increased perfu- proaches with functional imaging and histomorphometric sion as measured by PET. analyses are desirable to deliver information about phys- iologic effects caused by these therapeutic modalities. To

Received Aug. 26, 2005; revision accepted Nov. 21, 2005. provide a noninvasive, functional relevant measure of angio- For correspondence or reprints contact: Uwe Haberkorn, MD, PhD, genesis and to obtain functional data in animals and Department of Nuclear Medicine, University of Heidelberg, INF 400, Heidelberg, 69120, Germany. humans, MRI, CT, ultrasonography, optical imaging, and 15 E-mail: [email protected] PET have been proposed (9). Especially PET using H2 O

ANGIOSTATIN AND PERFUSION • Schmidt et al. 543 provides a quantitative measure of tissue perfusion—that is, in tumors with diameters between 10 and 13 mm (8–19 d after blood flow and blood volume in tumors (9). Furthermore, transplantation for WT-MH3924A and 37 and 57 d for hANG- recent data present evidence that 68Ga-DOTA-albumin MH3924A) were examined after intravenous bolus injection of 15 68 (DOTA is 1,4,7,10-tetraazacyclododecane-N,N9,N$,N%- 70–150 MBq H2 O in 0.3 mL and 5–10 MBq Ga-labeled 68 tetraacetic acid) PET can be used as a blood-pool marker in albumin [ Ga-DOTA-albumin] (same animals and session; the albumin PET scan was performed after the activity of H 15O vivo (10). The aim of this study was to evaluate the effects of 2 decayed). All PET studies were done as dynamic measurements hANG gene transfer on tumor growth and changes in perfu- 15 (20 · 3s,6· 10 s, 4 · 15 s, and 6 · 30 s for H2 O and 12 · 10 s, sion or blood volume relating in vivo measurements with PET 3 · 20 s, 4 · 30 s, and 10 · 60 s for 68Ga-DOTA-albumin) to immunohistochemistry and gene expression analysis. acquired in 2-dimensional (2D) mode using a matrix of 256 · 256 on an ECAT HR1 (Siemens/CTI) scanner (pixel size, 2.277 · MATERIALS AND METHODS 2.777 · 2.425 mm; transaxial resolution, 4.3 mm). A transmission Cell Culture and Generation of Recombinant Cells scan was done for 10 min before tracer administration with 3 rotating germanium pin sources to obtain cross sections for Cells were cultured (37C, 95% air/5% CO2) in RPMI 1640 medium (Gibco BRL) supplemented with 292 mg/L glutamine attenuation correction. After iterative reconstruction (ordered- and 20% fetal calf serum (FCS). subsets expectation maximization; 8 subsets, 16 iterations), dy- namic PET data were evaluated following definition of volumes of hANG Gene Transfer: Northern and Western Blots interest (VOIs) as described earlier using the PMOD software MH3924A cells were transfected with hANG (pGT60hAngio- package (13). Time–activity curves were created using VOIs, statin; InvivoGen) expression vectors. The hANG consists which consist of several regions of interest (ROIs) over the target of 385 amino acids, including the first 4 kringle structures. Stable area. Irregular ROIs were drawn manually. Perfusion studies using 15 cell lines were generated after lipofection of MH3924A cells by H2 O were evaluated using a 1-tissue-compartment model (13). hygromycin selection (425 mg/mL) for 4 wk. RNA was isolated For the input function, we used the mean value of the VOI data using TRIZOL reagent (Roche) and messenger RNA (mRNA) obtained from the heart. ROIs were defined at 3–6 s after bolus expression was assessed by Northern blots using radioactively injection in at least 3 consecutive slices (2.4-mm thickness) labeled probes delivered by digestion from corresponding plas- surrounded by 2 slices showing the heart in each direction. This mids. Secretion of hANG protein from selected clones into cell approach was based on the findings of Ohtake et al. showing that culture media was verified by Western blots. Conditioned medium the input function can be retrieved from the image data with was produced by culture of 106 transfected Morris hepatoma cells acceptable accuracy (14). The heart weight and heart volume of a in Optimem medium (Invitrogen) for 48 h. hANG protein was 250-g rat are 1.0 g and 1.2 mL, respectively (15). This results in an isolated from medium by incubation with L-lysine-Sepharose estimated recovery error of 10%, which can be neglected for the 68 matrix followed by 3 washing steps and elution with sodium purposes of this study. The Ga-DOTA-albumin uptake was dodecyl sulfate–polyacrylamide gel electrophoresis sample buffer. expressed as the mean standardized uptake value (SUV) between After gel electrophoresis, protein was immunoblotted on a poly- 10 and 15 min after injection when the time–activity curve reaches vinylidene fluoride membrane using polyclonal rabbit antihANG a plateau. The SUV is defined as: tissue concentration [Bq/g]/ (2 mg/mL; Oncogene) and the ECL Western Blotting (injected dose [Bq]/body weight [g]) (10). detection reagent (Amersham Biosciences). Tissue Preparation and Immunohistochemistry Measurement of Apoptotic or Proliferating Cells Rat tumors were fixed with immunohistochemistry zinc fixative Human umbilical vein endothelial cells (HUVECs) were (BD Biosciences), 4% paraformaldehyde (and embedded in par- isolated as described (11) and cultured in supplemented (4 ng/mL affin), or shock-frozen in liquid nitrogen-cooled isopentane and basic fibroblast growth factor, 20% FCS, 292 mg/L glutamine, kept in a freezer at 270C. Immunohistochemistry of the tumors 100,000 IU/L penicillin, 100 mg/L streptomycin) M199 medium. was performed on 6-mm cryostat sections according to procedures After 3 d of coculture with wild-type (WT)- or hANG-MH3924A as described (12,16). For immunohistochemical investigations, in a ‘‘without contact’’ system (pore size, 3 mm; BD Biosciences), monoclonal mouse antirat-CD31 (1:500; BD Bisosci- 3 proliferation of HUVECs was determined by methyl- H-thymidine ences), mouse anti–a-actin (1:100; Roche), mouse antiproliferating 3 ( H-TdR) (Amersham-Buchler) incorporation following treatment cell nuclear antigen (PCNA; 1:500; DakoCytomation), or mouse with 0.5 mol/L perchloric acid/1 mol/L NaOH and apoptosis was antirat CD11b (1:500; BD Biosciences) was applied. Negative measured in unfixed cells by YO-PRO-1 (Invitrogen) staining as controls were performed using control immunoglobulin or omis- described earlier (12). Coculture experiments were done in triplicate sion of the primary antibody. in 2 independent experiments. Apoptosis was measured on paraffin-embedded tumor cross sections by the terminal deoxynucleotidyl (TdT)- Animal Studies mediated dUTP nick end labeling (TUNEL) technique using a ACI and RNU (nude) rats were supplied by Charles River commercial kit (Oncor). Laboratories. All animal studies were performed in compliance with the national laws relating to the conduct of animal experi- mentation. The in vitro doubling time of the cell lines used was Morphometric Analysis 16.95 h for WT- and 17.4 h for hANG-MH3924A cells. After Immunoreactive cells were recorded by video camera (Olym- inoculation of WT or hANG cell suspensions (2 · 106 cells per pus HCC-3600 P high gain) and quantified using a computer- animal) subcutaneously into the thigh of ACI or RNU rats, tumor assisted image analysis system (VIBAM 0.0-VFG 1-frame grabber) growth was measured using calipers. Perfusion and blood volume as described earlier (16). Morphometric analyses were performed

544 THE JOURNAL OF NUCLEAR MEDICINE • Vol. 47 • No. 3 • March 2006 twice by 2 independent observers (interobserver variability, cytoplasm, small molecules) fraction, the proliferation of ,5%). HUVECs was measured by 3H-TdR incorporation into DNA. HUVECs cocultured with WT-MH3924A showed a 30% Gene Arrays increase of proliferation in comparison to HUVECs cultured Aliqots of WT- and hANG-MH3924A tumors obtained from alone. The proliferation rate of HUVECs decreased signif- the periphery were stored in TRIZOL to extract RNA according to the manufacturer’s specifications. RNA was purified using the icantly by 65% when cocultured with h-ANG-MH3924A in RNeasy kit (Qiagen Inc.) and quality was photometrically tested comparison with coculture with WT-MH3924A cells (Fig. 2A). by the 280:260 ratio and on agarose gel. Gene chip expression Furthermore, the percentage of apoptotic HUVECs cocul- arrays were performed using ‘‘The Rat Genome U34 Set’’ (Affimetrix tured with hANG-MH3924A increased 3.5-fold (P , 0.05) Inc.). The RNAs of hANG- (n 5 5) and WT-MH3924A (n 5 7) compared with coculture with WT-MH3924A (Fig. 2B). tumors were pooled and evaluated on separate gene chips as described earlier (16). The complete experiment was repeated twice, revealing a high correlation of signals in both experiments Inhibition of Tumor Growth and Proliferation and (hANG: r 5 0.96 [tumor], r 5 0.98 [cells]). Induction of Apoptosis in hANG-Overexpressing MH3924A Tumors In Vivo Statistical Methods After inoculation of tumor cell suspensions, we exam- Results are presented as mean 6 SEM. Statistical analyses ined growth of WT- and hANG-MH3924A tumors in rats, were performed by the Mann–Whitney U test or by the unpaired including histologic analysis of proliferation, necrosis, apo- Student t test using the SIGMASTAT program (Jandel Scientific). ptosis, and the content of oxidative stress inducing cells in P # 0.05 was considered statistically significant. these tumors.

RESULTS Inhibition of Proliferation and Induction of Apoptosis in Coculture Experiments with HUVECs Successful transfer of the hANG gene into MH3924A (hANG-MH3924A) and overexpression of the hANG gene were determined by Northern and Western blotting (Figs. 1A and 1B). A protein with a molecular weight of 56 kDa was identified in cell culture medium of hANG-overex- pressing cells (Fig. 1B). After treatment with perchloric acid/NaOH yielding an acid-insoluble (representing nucleic acids and ) and an acid-soluble (representing unbound radioactivity in acid- soluble molecules that are not in DNA and proteins—e.g.,

FIGURE 2. Proliferation and apoptosis of HUVECs cocultured FIGURE 1. Expression of hANG in clones of MH3924A. (A) with WT-MH3924A or hANG-MH3924A. (A) Proliferation (acid- Northern blot. (B) Western blot. Representative blots are shown. soluble and acid-insoluble fractions of thymidine uptake). (B) a-Tubulin or b-actin was used as internal control. bp 5 base Apoptosis. Values are shown as mean 1 SEM (*P , 0.05 vs. pairs. WT-MH3924A).

ANGIOSTATIN AND PERFUSION • Schmidt et al. 545 At 36 d after inoculation in ACI rats, in vivo growth of hANG-MH3924A tumors was reduced by 95% (Fig. 3). In vivo experiments, using a mixture of overexpressing cells (10% or 90%, respectively) and WT-MH3924, revealed that 36 d after inoculation tumor growth was inhibited by about 45% (90% hANG-MH3924A) and 20% (10% hANG- MH3924A) in comparison with the WT-MH3924A tumors. To exclude immunologic effects as possible causes for growth inhibition, we measured tumor size of hANG- and WT- MH3924A in athymic (nude) RNU rats. Eighteen days after inoculation, 90% suppression of tumor growth was noted in animals bearing hANG-overexpressing tumors (absolute value, 300 6 73 mL; n 5 4) in comparison with animals with WT-MH3924A (absolute value, 3,600 6 400 mL; n 5 5) (data not shown). Immunohistochemistry demonstrated that the prolifera- tion rate of hANG tumors (identified by PCNA staining) was significantly decreased (55%) (Fig. 4A). Additionally, the density of apoptotic cells increased (100%) in hANG- MH3924A tumors (Fig. 4B), whereas the density of p53 immunoreactive cells per millimeter2 was comparable in both groups (data not shown). In hANG-MH3924A, the necrotic area decreased significantly (P , 0.01) to 5.3% 6 0.8% in comparison with WT-MH3924A (21.9% 6 4.2%). Furthermore, in hANG-MH3924A tumors, the percentage of CD11b immunoreactive cells (8.4% 6 0.9%) decreased significantly (P , 0.001) in comparison with WT-MH3924A (29.6% 6 3.8%).

Increased Perfusion and Blood Volume in hANG-Overexpressing Hepatomas: PET and FIGURE 4. Histomorphometric analyses of apoptotic and proliferative cells (as indicated by TUNEL technique [A] or Immunohistomorphometry PCNA staining [B]) in WT-MH3924A and hANG-MH3924A Given that hANG has been described as a strong inhib- tumors. Cell density (nuclei/mm2) or percentage of positive itor of angiogenesis, we expected significant differences in cells was quantified by analyzing 5 ROIs (n 5 6 each group). tissue perfusion and blood volume in tumors overexpress- Values are shown as mean 1 SEM (***P , 0.001; **P , 0.01 vs. ing angiostatin. For tissue perfusion, dynamic PET mea- WT-MH3924A). 15 surements with H2 O were performed. Pharmacokinetic analysis of the PET data showed that, on the average, tumor perfusion, represented as the transport rate K1 (in mL · mL tissue21 · min21), increased significantly (P 5 0.001) in hANG-MH3924A (0.77 6 0.04) in comparison with WT- MH3924A (0.43 6 0.04; Fig. 5A). The k2 value did not differ in hANG-MH3924A (0.82 6 0.03) in comparison with WT-MH3924A (0.73 6 0.10; Fig. 5B). The fractional volume of distribution (DV) increased significantly in hANG-MH3924A (0.95 6 0.07; P 5 0.03) compared with WT-MH3924A (0.65 6 0.11; Fig. 5C), whereas the vascu- lar fraction (VB) did not differ significantly between WT- MH3924A (0.02 6 0.01) and hANG-MH3924A (0.05 6 0.02) tumors (Fig. 5D). To obtain noninvasive data about the perfused intravas- cular space of the tumor, dynamic blood-pool PET with 68Ga-DOTA-albumin was conducted and mean SUVs were FIGURE 3. Tumor growth of WT-MH3924A and hANG- calculated. Our blood-pool PET studies demonstrated a MH3924A tumors until 40 d after inoculation. Values are shown significantly increased blood volume of 28% in hANG- as mean 6 SEM (***P , 0.001; **P , 0.01; vs. WT-MH3924A). MH3924A (0.68 6 0.04) compared with WT-MH3924A

546 THE JOURNAL OF NUCLEAR MEDICINE • Vol. 47 • No. 3 • March 2006 15 FIGURE 5. H2 O PET. Tumor perfusion (K1,k2, DV, and VB values) measured in WT-MH3924A (n 5 7) and hANG- MH3924A (n 5 8) of ACI rats; 2D acqui- sition (1-tissue compartment model). hANG-expressing tumors showed a sig- nificantly increased K1 (A), whereas k2 remained unchanged (B). The fractional volume of distribution (DV) increased sig- nificantly in hANG-MH3924A (C), whereas the vascular fraction (VB) showed no dif- ference (D).

(0.49 6 0.07) (Fig. 6). To allow a comparison with the center (Fig. 7A). Using the generally accepted method to 15 H2 O study, tumor-to-heart ratios were calculated at 45–65 s determine microvessel density by measuring the CD31 im- after tracer administration with significantly (P 5 0.002) munoreactive area in angiogenic ‘‘hot spots’’ in the periphery increased values for hANG-MH3924A (0.8 6 0.19) as (17), we found a 40% increase in hANG-MH3924A tumors 15 compared with WT-MH3924A (0.47 6 0.12) for H2 O. (Fig. 7B). Microvessel density is 2–3 times higher in the The ratios for albumin showed similar changes (0.17 6 0.05 periphery in comparison with the center. In addition, the for hANG-MH3924A and 0.09 6 0.02 for WT-MH3924A mean number of blood vessels was significantly increased with P 5 0.007). in hANG- (330.14 6 28.80 per mm2) in comparison with The most commonly used vascular marker, a-actin, is ex- WT-MH3924A tumors (222.02 6 25.58 per mm2). pressed by mural cells of most arterioles and venules, but not in capillaries (9). In ACI rats we found that the a-actin Effects of hANG Gene Transfer on Expression of Other immunoreactive area is similar in hANG- and WT-MH3924A Genes In Vitro and In Vivo Gene Arrays tumors in the periphery as well as in the center; however, it To assess general effects of hANG gene transfer on is twice as high in the periphery in comparison with the tumor cells in vitro and in vivo, the gene expression pattern was studied using gene arrays (Table 1). Because transfer of selection markers such as the hygromycin gene and the selection procedure may cause changes in the genetic pro- gram, we analyzed the expression pattern in tumor cell lines as well as in explanted tumors of these cell lines. In general, we found changes in expression of genes related to apo- ptosis, signal transduction/stress, or metabolism/synthesis (Table 1).

DISCUSSION Gene transfer of antiangiogenic genes has been sug- gested as a new strategy of tumor therapy. Benefits of this therapeutic approach are the long-term expression of antiangiogenic genes, which is advantageous for the angio- static character of most antiangiogenic therapies and the genetic stability of endothelium. Another advantage is local production of therapeutic agents at the tumor site, thereby 68 FIGURE 6. Ga-DOTA-albumin PET: Tumor blood volume of preventing rapid degradation of therapeutic molecules in WT-MH3924A (n 5 5) and hANG-MH3924A (n 5 8) in ACI rats; 2D acquisition, SUVs were calculated. hANG-overexpressing plasma. Direct effects of hANG overexpression were ob- tumors demonstrated a significant increase in tumor blood served in our coculture experiments with HUVECs, result- volume in comparison with WT-MH3924A. ing in a decreased proliferation and increased apoptosis rate

ANGIOSTATIN AND PERFUSION • Schmidt et al. 547 The is not directly cytotoxic to tumor cells. However, it can increase tumor cell apoptosis and inhibit tumor growth by repressing endothelial cell prolif- eration/migration or by inducing endothelial cell apoptosis or both (21). Induction of apoptosis may be explained by diminished paracrine release of antiapoptotic factors and increased release of apoptotic factors from endothelial cells into tumor bed. To assess physiologic changes in tumor perfusion caused 15 68 by angiostatin, PET studies with H2 O and Ga-DOTA- 15 albumin were performed. H2 O is perfectly suited for examining tumor tissue perfusion, because it is an inert and freely diffusible molecule not participating in metabolic processes at least during PET measurements (22,23), whereas 68Ga-DOTA-albumin, a tracer of the blood vol- ume, persists in the intravascular space during PET studies. 15 68 Investigations with H2 O and Ga-DOTA-albumin in hANG-expressing tumors revealed increased tissue perfu- sion as well as increased blood volumes in comparison with WT-MH3924A (Fig. 5; Fig. 6). The higher tumor-to-heart 15 ratios for H2 O as compared with albumin are explained by the different distribution of these tracers. The pharma- cokinetic model assumes that K1 is in agreement with blood flow if the extraction rate is high. The model assumes that water is a completely and freely diffusible molecule with 100% extraction from circulation (24). In this model, the 15 fractional DV is the proportion of the ROI in which H2 O is distributed (25). For example, a value of 2 means that during equilibrium the radioactive concentration in tissue is twice as high compared with that in blood. Because WT- FIGURE 7. Immunohistomorphometric analysis of vasculari- and hANG-expressing tumors did not differ in diameter, zation in Morris hepatomas. WT-MH3924A and hANG- changes in tissue perfusion or blood volumes are not due to MH3924A were stained with a-actin or CD31 antibodies (each partial-volume effects. In addition, our PET data are in ac- group, n 5 6). Control immunoglobulin (data not shown) was used as negative controls. Immunoreactive areas were quan- cordance with our histologic findings that microvessel den- tified by analyzing 5 hot spots per tumor. Center 5 center of sity is also significantly increased in hANG-overexpressing tumor. Values are shown as mean 1 SEM (**P , 0.01 vs. WT- hepatomas, whereas the fraction of larger vessels (e.g., arte- MH3924A). rioles) containing a-actin immunoreactive smooth muscle cells was comparable between WT- and hANG-MH3924A of HUVECs compared with coculture with WT-MH3942A tumors (Figs. 7A and 7B). Similar data were obtained with cells (Figs. 2A and 2B). Similar results have been demon- , where evidence for a dose-dependent biphasic 15 strated in HUVECs cocultured with a stable ANG-express- change of tumor perfusion measured by H2 O PET was ing hepatocellular carcinoma cell line, PLC/PRF/5 (18). found in a phase I clinical trial (26). Lower doses of endo- Furthermore, these results are in concert with recent reports statin resulted in an increase of tumor blood flow, whereas a showing that angiostatin inhibits cell proliferation and decrease of tumor blood flow was noted when intermediate induces apoptosis in bovine aortic endothelial cells and or high doses were applied. Improvement of perfusion and HUVECs (19). dose-dependent response on antiangiogenic agents has been We observed a significant inhibition of tumor growth in explained by the hypothesis of a decrease in interstitial fluid animals bearing hANG-MH3924A tumors (Fig. 3). Similar pressure or normalization of tumor vasculature followed by results of tumor growth suppression were noted in mice diminished hypoxia (27). Although it may be speculated bearing ANG-expressing hepatocellular carcinoma (18)or that a normalization of tumor vasculature stimulates tumor because of application of recombinant adeno-associated proliferation, animal and clinical studies to date show no virus encoding ANG plus endostatin using an ovarian acceleration of tumor growth during antiangiogenic mono- cancer cell model (20). Our immunohistochemical analyses therapy despite tumor vessel normalization (27). indicated decreased proliferation together with an increased We also found that the necrotic area of tumor sections apoptosis rate in these tumors (Figs. 4A and 4B), providing was diminished in hANG-MH3924A. This corresponds to additional evidence for antitumoral effects of angiostatin. the higher microvessel density and the improvement of

548 THE JOURNAL OF NUCLEAR MEDICINE • Vol. 47 • No. 3 • March 2006 TABLE 1 Changes in Gene Expression in Genetically Modified hANG-Expressing Hepatomas

Gene Tumor ratio Cell ratio

Matrix associated -X 2.66 0.31 Apoptosis associated c-myc, exon-2 5.41 1.22 Cyclin-dependent kinase 2-b 4.36 1.33 Rab3 GDP/GTP exchange protein 3.03 4.58 A2b-adenosine receptor 2.63 0.84 Cyclin D3 2.29 0.93 Retinoic acid receptor a1 2.04 0.82 fos-related antigen (Fra-1) 0.23 0.87 Signal transduction and stress g-Glutamyltranspeptidase 9.96 0.79 Ras-related mRNA rab4 8.51 0.63 a-B-crystallin 5.80 0.36 Angiotensin receptor (AT1) 4.89 0.92 b-1 subunit of Na,K-adenosine triphosphatase 4.08 0.94 Syntaxin 2 3.94 0.94 Cysteine dioxygenase (CDO) 3.64 1.09 SH3-containing protein p4015 3.28 0.41 Phosphatidylinositol 4-kinase 2.97 0.93 Rhodanese 2.96 0.66 Glutathione reductase 2.86 1.57 CLC-7 chloride channel protein 2.75 0.53 Protein kinase C type I 2.74 1.20 Macrophage-stimulating protein 2.71 0.78 Rolipram-insensitive phosphordiesterase type 7 (RPDE7-1) 2.49 0.20 b-B3-2-crystallin 2.47 0.72 Thromboxane A2 receptor 2.26 1.09 Cystathionine g- (CSE) 2.19 0.87 Adenylyl cyclase type V 2.03 0.97 Macrophage inflammatory protein-1a 0.39 0.31 Interleukin-1a 0.38 0.42 Inducible nitric oxide synthase (NOS) 0.33 0.47 Cytochrome P450 IIA2 0.30 1.07 Insulin-like growth factor I 0.30 0.69 Metabolism/synthesis UDP 5.07 1.31 P2X 4.50 0.85 Hepatic microsomal UDP-glucuronosyltransferase (UDPGT) 3.37 1.08 WAP 4-disulfide core domain protein (ps20) 3.27 0.46 (B-GP1) 2.68 1.25 -specific UDP-glucuronosyltransferase pyruvate dehydrogenase 2.58 0.74 Phosphatase isoenzyme 1 0.34 0.58 Miscellaneous a-Actinin-2–associated LIM protein 7.72 0.28 Class Ib RT1 7.25 0.41 MHC class II H-b 6.74 1.26 Factor X 5.56 0.19 Polyprotein 1-/bikunin 4.23 0.52 a-1A-adrenergic receptor 3.94 1.94 Memory-related gene-7 3.76 2.41 Nonselective-type endothelin receptor 3.50 0.34 Cell adhesion regulator (CAR1) 3.10 1.11 e1 globin hereditary hemochromatosis protein 3.06 0.93 Homolog (RT1-CAFe) 2.96 0.98 CD5 2.89 0.74 Melanocyte-specific gene 1 protein (msg1) 2.47 0.82 Chromosomal protein HMG2 2.46 0.40

Ratios of modified vs. WT tumors and cells are shown.

ANGIOSTATIN AND PERFUSION • Schmidt et al. 549 perfusion in these tumors. Therefore, our results imply a Tenascin-X regulates the development of blood vessels better exchange of oxygen and nutrients as well as an im- during vasculogenesis and angiogenesis (33). The observed proved removal of toxic metabolites from tumor tissue in upregulation of tenascin-X gene expression may be one of hANG-overexpressing tumors followed by a decrease of the causes for the increase in vascularization or perfusion hypoxia but without enhancing tumor growth. Changes in found in hANG-MH3924A tumors. hypoxia may have multiple consequences for the tumor bi- The increase in the percentage of apoptotic cells in ology, such as promotion of genomic instability (28) and de- hANG-MH3924A tumors may be attributed to an induction creased secretion of tumor growth factors. The consequences of several proapoptotic genes in these tumors, such as cyclin- of hypoxia probably depend on the tumor gene expression dependent kinase 2 (cdk2) and c-myc, which together have level and microenvironment. Additionally, survival of cer- been shown to be critical in the decision between quies- tain tumors cells is based on the same angiogenic growth cence and apoptosis pathways, sharing regulatory machin- factors as endothelial cells. This implies that reduction of ery with cell cycle control mechanisms (34). Also, the growth factor levels results in inhibition of tumor and A2b-adenosine receptor (P1 receptor for adenosine) is endothelial cell growth inducing tumor regression (27). involved in apoptosis induction in different cell types With regard to increased vascularization and perfusion in (35). Adenosine receptors bind adenosine, which is an tumors coexisting with significant growth inhibition, vas- ubiquitous nucleoside present in all body cells being cularization in transfected tumors may not be the limiting released from metabolically active or stressed cells (36). factor responsible for tumor growth inhibition. In this con- Modulation of stress-relevant parameters in hANG- text, little attention has been paid to the permeability of MH3924A tumors is evidenced by the increased expression blood vessels in tumors. Although the leakiness of tumor of stress-related genes, such as genes of the cysteine/ vessels is well documented in human cancer and in exper- glutathione-associated or sulfur-delivery metabolism (g- imental tumor models, the mechanism is poorly understood glutamyltranspeptidase, crystalline, cysteine dioxygenase (29). It has been shown recently that treatment with ANG [CDO], rhodanese, glutathione reductase, cystathionine for 5 d reduced permeability of tumor blood vessels (30). g-lyase [CSE]). The increased expression of CSE, CDO A reduction of vessel permeability associated with in- (a critical regulator of cysteine concentration), g-glutamyl- creased microvessel density in hANG-MH3924A tumors transpeptidase (which limits glutathione synthesis by pre- may result in an improved functional flow and decreased venting cysteine generation from extracellular glutathione), interstitial fluid pressure preventing hypoxia, a main stim- and glutathione reductase indicates a modulation of the ulus of angiogenesis. thiol/cysteine/glutathione metabolism by hANG, with a The permeability of vessels may be increased by inflam- possible impact on the redox state of the tumor-associated matory mediators and leakage may be related to the cells or apoptosis/proliferation. In this context, overexpres- migration of leukocytes through vessel walls into tissue sion of CSE has been shown recently to inhibit cell (29), producing multiple factors for angiogenesis or in- proliferation and cell growth (37). Furthermore, rhodanese flammation (31). Recently, it was shown that infiltrating is believed to function as a sulfur-delivery protein (38) CD11(1) cells were the major source of oxidative stress being involved in detoxification of intramitochondrial oxy- and their cellular infiltration into the tissue corresponded to gen free radicals (39). Additionally, we found other stress/ an enhanced tumor growth (32). Given that the percentage signal transduction genes to be elevated in hANG-MH3924A of CD11b(1) cells was significantly diminished in hANG- tumors, such as the angiotensin II type 1 (AT1) receptor, MH3924A tumors, this finding is compatible with a re- which is upregulated in cancer through systemic oxidative duced permeability of the vessels and is interpreted as a stress and hypoxia mechanisms (40). decrease of oxidative stress, resulting in diminished growth Finally, we found induction of the expression of several stimulation. genes coding for involved in metabolism/synthe- To assess general escape phenomena of the cells or sis of cholesterol/fatty acids/triglycerides, possibly indicat- tumors by changes in the expression pattern, gene array ing a high energy expenditure or cell transformation. analyses were performed. Gene arrays of the genetically modified cells reveal that overexpression of hANG affects expression of multiple genes. The different gene expression CONCLUSION in hANG tumors as compared with hANG cells indicates Transfer of the hANG gene results in inhibition of tumor differences between the in vitro situation as opposed to the growth, which may be due to multiple factors such as in vivo situation, where the microenvironment—that is, a decreased proliferation, increased apoptosis, and general mixture of cells (hepatoma cells, cells of the vessel wall or changes in gene expression. At the genetic level, we of the connective tissue), metabolic interaction of cells, observed that hANG induces changes in the expression of oxygen supply, and so forth—includes potential additional multiple genes related mainly to apoptosis, signal trans- factors. In genetically modified tumors, we found signifi- duction or stress, and metabolism. Because these changes cant changes in a variety of genes related to apoptosis, in gene expression pattern are predominantly observed in signal transduction/stress, and metabolism/synthesis. tumor tissue and not in vitro, they represent reactions of the

550 THE JOURNAL OF NUCLEAR MEDICINE • Vol. 47 • No. 3 • March 2006 tumor to its changing microenvironment. Furthermore, an 15. Sharp PE, LaRegina MC. The Laboratory Rat. Boca Raton, FL: CRC Press; 1998. increase in perfusion and blood volume was measured in 16. Schmidt K, Hoffend J, Altmann A, et al. Transfer of the sFLT-1 gene in Morris 15 68 PET studies with H2 O and Ga-DOTA-albumin, corre- hepatoma results in decreased growth and perfusion and induction of genes sponding to increased microvessel density and a decrease in associated with stress response. Clin Cancer Res. 2005;11:2132–2140. 17. Fox SB, Harris AL. Diagnostic and prognostic significance of tumor angiogen- necrosis. Therefore, PET studies with these 2 tracers seem esis. In: Fan TPD, Kohn E, eds. The New Angiotherapy. Totowa, NJ: Humana to be a sensitive noninvasive tool for the characterization of Press Inc.; 2002:151-176. these changes induced by angiostatin in vivo. 18. Ishikawa H, Nakao K, Matsumoto K, et al. Antiangiogenic gene therapy for hepatocellular carcinoma using angiostatin gene. Hepatology. 2003;37:696–704. 19. Hanford HA, Wong CA, Kassan H, et al. Angiostatin(4.5)-mediated apoptosis of vascular endothelial cells. Cancer Res. 2003;63:4275–4280. ACKNOWLEDGMENTS 20. Isayeva T, Ren C, Ponnazhagan S. Recombinant adeno-associated virus The authors thank Helmut Eskerski, Karin Leotta, and 2-mediated antiangiogenic prevention in a mouse model of intraperitoneal ovarian cancer. Clin Cancer Res. 2005;11:1342–1347. Uschi Schierbaum for their help in performing the animal 21. Folkman J. Angiogenesis and apoptosis. Semin Cancer Biol. 2003;13:159–167. experiments as well as Silke Vorwald and Ulrike Traut for 22. Bacharach SL, Libutti SK, Carrasquillo JA. Measuring tumor blood flow with 15 technical assistance in histology. Financial support was H2 O: practical considerations. Nucl Med Biol. 2000;27:671–676. 23. Bruehlmeier M, Roelcke U, Schubiger PA, Ametamey SM. Assessment of from the Wilhelm Sander Stiftung (1999.085.1) and the hypoxia and perfusion in human brain tumors using PET with 18F-fluoromiso- 15 Tumorzentrum Heidelberg/Mannheim, the Bequest of nidazole and O-H2O. J Nucl Med. 2004;45:1851–1859. Herbert Dauss, as well as the Foundation for Cancer and 24. Anderson H, Price P. Clinical measurement of blood flow in tumours using positron emission tomography: a review. Nucl Med Commun. 2002;23:131–138. Scarlatina Research. 25. Anderson H, Yap JT, Wells P, et al. Measurement of renal tumour and normal tissue perfusion using positron emission tomography in a phase II clinical trial of razoxane. Br J Cancer. 2003;89:262–267. REFERENCES 26. Herbst RS, Mullani NA, Davis DW, et al. Development of biologic markers of response and assessment of antiangiogenic activity in a clinical trial of human 1. Folkman J. Tumor angiogenesis: therapeutic implications. N Engl J Med. recombinant endostatin. J Clin Oncol. 2002;20:3804–3814. 1971;285:1182–1186. 27. Jain RK. Normalization of tumor vasculature: an emerging concept in 2. Bergers G, Benjamin LE. Tumorigenesis and the angiogenic switch. Nat Rev antiangiogenic therapy. Science. 2005;307:58–62. Cancer. 2003;3:401–410. 28. Nelson DA, Tan TT, Rabson AB, Anderson D, Degenhardt K, White E. Hypoxia 3. Kerbel RS. Inhibition of tumor angiogenesis as a strategy to circumvent acquired and defective apoptosis drive genomic instability and tumorigenesis. Genes Dev. resistance to anti-cancer therapeutic agents. Bioessays. 1991;13:31–36. 2004;18:2095–2107. 4. O’Reilly MS, Holmgren L, Shing Y, et al. Angiostatin: a novel angiogenesis 29. McDonald DM, Baluk P. Significance of blood vessel leakiness in cancer. inhibitor that mediates the suppression of metastases by a Lewis lung carcinoma. Cancer Res. 2002;62:5381–5385. Cell. 1994;79:315–328. 30. Satchi-Fainaro R, Mamluk R, Wang L, et al. Inhibition of vessel permeability 5. Ji WR, Castellino FJ, Chang Y, et al. Characterization of kringle domains of by TNP-470 and its polymer conjugate, caplostatin. Cancer Cell. 2005;7: angiostatin as antagonists of endothelial cell migration, an important process in 251–261. angiogenesis. FASEB J. 1998;12:1731–1738. 31. Carmeliet P. Angiogenesis in health and disease. Nat Med. 2003;9:653–660. 6. te Velde EA, Vogten JM, Gebbink MF, van Gorp JM, Voest EE, Borel Rinkes IH. 32. Sluyter R, Halliday GM. Enhanced tumor growth in UV-irradiated skin is Enhanced antitumour efficacy by combining conventional with associated with an influx of inflammatory cells into the epidermis. Carcinogen- angiostatin or endostatin in a liver metastasis model. Br J Surg. 2002;89:1302–1309. esis. 2000;21:1801–1807. 7. Soff GA. Angiostatin and angiostatin-related proteins. Cancer Metastasis Rev. 33. Ikuta T, Ariga H, Matsumoto K. tenascin-X in combination 2000;19:97–107. with vascular endothelial growth factor B enhances endothelial cell proliferation. 8. Dong Z, Kumar R, Yang X, Fidler IJ. Macrophage-derived metalloelastase is Genes Cells. 2000;5:913–927. responsible for the generation of angiostatin in Lewis lung carcinoma. Cell. 34. Helbing CC, Wellington CL, Gogela-Spehar M, Cheng T, Pinchbeck GG, 1997;88:801–810. Johnston RN. Quiescence versus apoptosis: Myc abundance determines pathway 9. McDonald DM, Choyke PL. Imaging of angiogenesis: from microscope to of exit from the cell cycle. Oncogene. 1998;17:1491–1501. clinic. Nat Med. 2003;9:713–725. 35. Le Poole IC, Yang F, Brown TL, et al. Altered gene expression in melanocytes 10. Hoffend J, Mier W, Schuhmacher J, et al. Gallium-68-DOTA-albumin as a PET exposed to 4-tertiary butyl phenol (4-TBP): upregulation of the A2b adenosine blood-pool marker: experimental evaluation in vivo. Nucl Med Biol. receptor 1. J Invest Dermatol. 1999;113:725–731. 2005;32:287–292. 36. Ohana G, Bar-Yehuda S, Barer F, Fishman P. Differential effect of adenosine on 11. Jaffe EA, Nachman RL, Becker CG, Minick CR. Culture of human endothelial tumor and normal cell growth: focus on the A3 adenosine receptor. J Cell cells derived from umbilical veins. identification by morphologic and immuno- Physiol. 2001;186:19–23. logic criteria. J Clin Invest. 1973;52:2745–2756. 37. Yang G, Cao K, Wu L, Wang R. Cystathionine gamma-lyase overexpression 12. Haberkorn U, Bellemann ME, Brix G, et al. Apoptosis and changes in glucose inhibits cell proliferation via a H2S-dependent modulation of ERK1/2 transport early after treatment of Morris hepatoma with gemcitabine. Eur J Nucl phosphorylation and p21Cip/WAK-1. J Biol Chem. 2004;279:49199–49205. Med. 2001;28:418–425. 38. Ogasawara Y, Lacourciere G, Stadtman TC. Formation of a selenium-substituted 13. Dimitrakopoulou-Strauss A, Strauss LG, Burger C. Quantitative PET studies in rhodanese by reaction with selenite and glutathione: possible role of a protein pretreated melanoma patients: a comparison of 6-[18F]fluoro-L-dopa with 18F- perselenide in a selenium delivery system. Proc Natl Acad Sci U S A. FDG and 15O-water using compartment and noncompartment analysis. J Nucl 2001;98:9494–9498. Med. 2001;42:248–256. 39. Nandi DL, Horowitz PM, Westley J. Rhodanese as a thioredoxin oxidase. Int J 14. Ohtake T, Kosaka N, Watanabe T, et al. Noninvasive method to obtain input Biochem Cell Biol. 2000;32:465–473. function for measuring tissue glucose utilization of thoracic and abdominal 40. Smith GR, Missailidis S. Cancer, inflammation and the AT1 and AT2 receptors. organs. J Nucl Med. 1991;32:1432–1438. J Inflamm (Lond). 2004;1:3–15.

ANGIOSTATIN AND PERFUSION • Schmidt et al. 551